Distinguishing between recruitment and spread of silent chromatin structures in Saccharomyces cerevisiae

The formation of heterochromatin at HML, HMR, and telomeres in Saccharomyces cerevisiae involves two main steps: the recruitment of Sir proteins to silencers and their spread throughout the silenced domain. We developed a method to study these two processes at single basepair resolution. Using a fusion protein between the heterochromatin protein Sir3 and the nonsite-specific bacterial adenine methyltransferase M.EcoGII, we mapped sites of Sir3–chromatin interactions genome-wide using long-read Nanopore sequencing to detect adenines methylated by the fusion protein and by ChIP-seq to map the distribution of Sir3–M.EcoGII. A silencing-deficient mutant of Sir3 lacking its Bromo-Adjacent Homology (BAH) domain, sir3-bah∆, was still recruited to HML, HMR, and telomeres. However, in the absence of the BAH domain, it was unable to spread away from those recruitment sites. Overexpression of Sir3 did not lead to further spreading at HML, HMR, and most telomeres. A few exceptional telomeres, like 6R, exhibited a small amount of Sir3 spreading, suggesting that boundaries at telomeres responded variably to Sir3-M.EcoGII overexpression. Finally, by using a temperature-sensitive allele of SIR3 fused to M.ECOGII, we tracked the positions first methylated after induction and found that repression of genes at HML and HMR began before Sir3 occupied the entire locus.


Introduction
Cells have an interest in coordinating the expression of genes: It allows them to turn sets of genes on and off in response to various stimuli or ensure certain genes are always expressed or always repressed to create and maintain cell identity. There are multiple ways to coordinate transcription, including shared binding sites for activators or repressors in promoters, nuclear compartmentalization, and creation of large domains like heterochromatin. The establishment of coordinated, stable blocs of gene expression, such as heterochromatin, can be broken down into two main steps: nucleation, which involves the recruitment of chromatin-modifying factors, followed by the expansion of these chromatin-modifying factors beyond recruitment sites in an ill-defined process known as spreading.
An impressive example of the concepts of nucleation and spread is inactivation of the X chromosome in female mammals (reviewed in Galupa and Heard, 2018). Nucleation begins with the transcription of the noncoding RNA Xist from one of the two X chromosomes, which recruits heterochromatin

Results
The Sir3-M.EcoGII fusion protein strongly and specifically methylated HML and HMR To achieve a higher-resolution method for assessing SIR complex binding, we made a fusion protein between Sir3 and M.EcoGII ( Figure 1A), a nonsite-specific bacterial N6-methyladenosine methyltransferase (Murray et al., 2018;Woodcock et al., 2020). In principle, wherever Sir3 binds chromatin, even transiently, M.EcoGII has the opportunity to methylate nearby accessible adenines to make m 6 A. S. cerevisiae has no endogenous DNA methylation and no demethylases, allowing us to attribute m 6 A only to the activity of the fusion protein, reflecting where it resides as well as where it has been. M.EcoGII has no specific recognition sequence, which should provide more resolution than methyltransferases like E. coli Dam, which has a four basepair recognition site. The positions of Sir3-M.EcoGII can be determined conventionally by immunoprecipitation with an antibody against m 6 A followed by sequencing the precipitated DNA using Illumina sequencing. The more powerful implementation would come from distinguishing between individual m 6 A bases and unmodified A bases using long-read Nanopore sequencing (Xu and Seki, 2020).
To test this concept, we first assessed the silencing ability of Sir3-M.EcoGII by measuring mRNA expression of HMLα2 and HMRa1. Strains expressing Sir3-M.EcoGII could silence HML and HMR as well as wild-type strains compared to sir3∆ strains that displayed a full loss of silencing ( Figure 1B). Thus, the fusion protein retained full function of Sir3.
To compare the binding profile of Sir3-M.EcoGII to the distribution of methylation it produced, we performed DNA immunoprecipitation and Illumina sequencing (DIP-seq) using an antibody that specifically recognizes m 6 A alongside ChIP-seq for a V5 epitope-tagged Sir3-M.EcoGII. ChIP-seq revealed strong Sir3-M.EcoGII occupancy over the E and I silencers at HML and over the E silencer at HMR with weaker but consistent signal above background between the two silencers ( Figure 1C, top row). Compared to ChIP-seq, methylation measured by DIP-seq had a stronger signal over the entirety of HML and HMR and a broader signal that extended beyond the silencers ( Figure 1C, Figure 1-figure supplement 1). The methylation over HML and HMR was from the fusion protein Sir3-M.EcoGII, as neither a strain without M.EcoGII nor a strain expressing unfused M.EcoGII from the SIR3 promoter showed appreciable DIP-seq signal ( Figure 1C, Figure 1-figure supplement 1).
Methylation by Sir3-M.EcoGII measured by Nanopore sequencing agreed well with DIP-seq, showing strong methylation over HML and HMR with little background methylation outside of these regions ( Figure 1D, Figure 1-figure supplement 2A). Fusions of Sir2 and Sir4 with M.EcoGII were also fully silencing competent ( Figure 1B) and produced methylation signals that matched Sir3-M. EcoGII at HML and HMR ( Figure 1D), suggesting that all three members of the SIR complex were equally distributed, as expected. In addition to the aggregate methylation signal (% of reads methylated at each position), we analyzed methylation of single adenines on single reads across HML and HMR ( Figure 1E, Figure 1-figure supplement 2B). Sir3-M.EcoGII methylated most strongly near the E and I silencers and at the HMLα1/α2 promoter, with lower, but significant, methylation between these sites ( Figure 1E, Figure 1-figure supplement 2B). Analysis of single reads revealed a periodicity of methylation across HML and HMR ( Figure 1E, Figure 1-figure supplement 2B). These small regions of higher methylation corresponded to linker regions between nucleosomes (Figure 1figure supplement 3). Sir3-M.EcoGII did methylate within nucleosome-occupied regions at HML and HMR but at a lower frequency (Figure 1-figure supplement 3), consistent with in vitro studies that use methylation by M.EcoGII or another nonspecific adenine methyltransferase, Hia5, as a measurement of chromatin accessibility (Abdulhay et al., 2020;Brady et al., 2021;Shipony et al., 2020;Stergachis et al., 2020).
Nucleosome binding was required for spreading, but not recruitment, of Sir3 The recruitment of the SIR complex to silencers is sequence specific. Rap1, Abf1, and ORC bind at these recruitment sites and recruit the Sir proteins directly through protein-protein interactions. In contrast, SIR complex binding away from recruitment sites is sequence independent and instead relies on interactions with nucleosomes and among the Sir proteins themselves.  Figure 1-figure supplement 1).
The online version of this article includes the following figure supplement(s) for figure 1:   We therefore hypothesized that the interaction between Sir3 and nucleosomes would not be required for nucleation at recruitment sites but would be required for binding outside of those recruitment sites. Sir3 has multiple recognized domains ( Figure 2A): The Bromo-Adjacent Homology (BAH) domain, which interacts with nucleosomes (Armache et al., 2011;Buchberger et al., 2008;Norris et al., 2008;Onishi et al., 2007;Rudner et al., 2005;Sampath et al., 2009), the AAA + domain, which interacts with Sir4 (Ehrentraut et al., 2011;King et al., 2006;Samel et al., 2017), and the winged helix (wH) domain, which allows for homodimerization (King et al., 2006;Liaw and Lustig, 2006;Oppikofer et al., 2013). We deleted the BAH domain of Sir3-M.EcoGII (bah∆), which largely abrogates Sir3-nucleosome interactions in vitro (Buchberger et al., 2008;Onishi et al., 2007;Sampath et al., 2009). Previous studies established that deletion of the BAH domain causes a phenotypic loss of silencing, likely due to a loss of interaction with nucleosomes (Buchberger et al., 2008;Gotta et al., 1998;Onishi et al., 2007), but did not characterize its binding at regions of heterochromatin. Importantly for what follows, deletion of the BAH domain did not destabilize Sir3 ( Figure 2B). We also confirmed that sir3-bah∆-M.ECOGII strains displayed a loss of silencing, but found that the loss was more severe at HML than at HMR ( Figure 2C).
Despite the loss of silencing at HML and HMR in the sir3-bah∆-M.ECOGII strain, there was still detectable methylation across the two loci ( Figure 2D, E, Figure 2-figure supplement 1A). At the aggregate level, sir3-bah∆-M.EcoGII methylated silencers at HML and HMR at the same level as Sir3-M.EcoGII but showed decreased methylation between them ( Figure 2D). Analysis of single reads spanning HML and HMR in the bah∆ mutant showed similarly strong levels of methylation at silencers, and revealed strong methylation both at the promoters of HML and HMR and the recognition site for the HO endonuclease ( Figure 2E). In contrast, little methylation was seen over gene bodies between these sites ( Figure 2E).
The ability of Nanopore to sequence long reads also allowed mapping and analysis of Sir3 occupancy on the repetitive and highly homologous telomeres, an issue with short-read Illumina sequencing. In addition to methylating HML and HMR (Figure 1), Sir3-M.EcoGII strongly methylated telomeres at TG repeats and X elements ( Figure 2F , and the periodicity of methylation was apparent on single reads as well ( Figure 2G), likely corresponding to more-accessible linker regions. The loss of binding outside of recruitment sites of sir3-bah∆-M.EcoGII was more striking at telomeres than HML and HMR, where methylation by the bah∆ mutant matched wild-type levels at TG and X repeats but dropped off steeply centromere-proximal to the X elements ( Figure 2F The results at telomeres, supported by the data at HML and HMR, suggested that the nucleosomebinding activity of Sir3 was required for Sir3 to spread away from recruitment sites, but not for its initial recruitment. SIR3 expression level did not limit its spread from recruitment sites In addition to understanding what enables Sir3 spreading, we were also interested in what limits its spread. One common feature of heterochromatin proteins is that their activity is dose dependent: lowered expression causes loss of heterochromatin whereas elevated expression can cause silencing of genes near heterochromatin (Henikoff, 1996;Locke et al., 1988). Indeed, it was previously reported that overexpression of Sir3 results in its spread beyond wild-type boundaries, accompanied by repression of genes in those extended regions (Hecht et al., 1996;Hocher et al., 2018;Ng et al., 2003;Renauld et al., 1993;Strahl-Bolsinger et al., 1997). To provide an independent test of those conclusions, we tested whether the expression level of Sir3 limited how far it could spread beyond recruitment sites at HML, HMR, and telomeres. mRNA levels of SIR3-M.ECOGII on a multicopy 2μ plasmid was tenfold higher than from the chromosomal SIR3-M.ECOGII locus in a strain carrying an empty 2μ plasmid ( Figure 3A). We confirmed the overexpression of SIR3-M.ECOGII in these strains by protein immunoblotting ( Figure 3B). Overexpression of SIR3-M.ECOGII had no effect on silencing at HML and HMR ( Figure 3C).
Overexpression of SIR3-M.ECOGII had little effect on the boundaries of methylation at HML and HMR. Strains overexpressing SIR3-M.ECOGII displayed increased methylation over both loci and between HML and telomere 3 L, but no new sites of methylation appeared outside the bounds of strains expressing only one copy of SIR3-M.ECOGII ( Figure 3D). The changes in methylation upon overexpression of Sir3-M.EcoGII matched changes in occupancy of the fusion protein by ChIP-seq,  There was some increase in methylation over the promoters of two genes closest to HML and HMR (CHA1 and OCA4, respectively), but it did not result in any change in the level of their expression ( Figure 3E). Surprisingly, the results at telomeres were qualitatively similar but revealed three categories of effects. Some telomeres showed a large increase in the amount of methylation upon overexpression of SIR3-M.ECOGII with a small extension of binding farther into the chromosome ( Figure 3F, top row, Figure 3-figure supplement 2). Some telomeres showed a modest increase in the amount methylation upon overexpression of SIR3-M.ECOGII with little, if any, extension of range ( Figure 3F, middle row, Figure 3-figure supplement 2). Finally, some telomeres showed no appreciable change in methylation ( Figure 3F, bottom row, Figure 3-figure supplement 2). As at HML and HMR, the changes in methylation seen with Nanopore sequencing matched with changes in occupancy measured by ChIP-seq ( Figure 3-figure supplement 1). Only three of the 32 telomeres, including the paradigmatic telomere 6 R from earlier studies (Hecht et al., 1996;Renauld et al., 1993;Strahl-Bolsinger et al., 1997), showed convincing spread of methylation to new sites compared to telomeres in strains expressing one copy of SIR3-M.ECOGII ( Figure 3F, Figure 3-figure supplement 2). Therefore, the expression level of SIR3 was not a universal limiting factor in heterochromatin spread. These results imply the existence of other chromatin features that create boundaries for Sir3 spreading.

Repression of HML and HMR preceded heterochromatin maturation
To evaluate the dynamics of Sir3 recruitment and spreading during the establishment of heterochromatin over time, we used a temperature-sensitive allele of SIR3, sir3-8, fused to M.ECOGII and took samples at various time points for Nanopore sequencing after switching from the restrictive (37°C) to the permissive (25°C) temperature ( Figure 4A). In agreement with previous studies (Stone et al., 2000), growth at 37°C caused lower protein levels of sir3-8 ( Figure 4B). Over the course of 150 min, the protein levels of sir3-8 slowly increased to 63% of the level in constitutive 25°C growth conditions ( Figure 4B). In cells grown at 37°C, sir3-8-M.ECOGII did not methylate HML and HMR, but when grown constitutively at 25°C, there was strong methylation over both loci ( Figure 4C, Figure 4figure supplement 1).
Over a 90-min time course, methylation increased only over the silencers and promoters of HML and HMR ( Figure 4D, Figure 4-figure supplement 2, solid lines). Methylation at the promoter of HML during the time course and in the sir3-bah∆ mutant was expected, as it contains a Rap1-binding site, and Rap1 interacts directly with Sir3 and Sir4. However, methylation at the promoter of HMR at these early time points and in the sir3-bah∆ mutant was a surprise, as there is not a known SIR complex-interacting protein that binds at the promoter. In the absence of Sir3, Sir4 is still bound at silencers (Goodnight and Rine, 2020;Hoppe et al., 2002;Rusché et al., 2002), so the faster recruitment at these sites, and perhaps the promoters as well, might be due to the interaction of Sir3 with Sir4 and Rap1.
By 90 min (~1 cell division), methylation at no position reached the level found in cells constitutively grown at 25°C -the level of methylation of mature, stable heterochromatin ( Figure 4D, Figure 4-figure supplement 2, dotted line). Strikingly, even by 30 min after the temperature shift, when methylation was just rising above background at silencers, partial repression of HML and HMR 4 R, and 11 R) from long-read Nanopore sequencing of the same strains as D. Shown are 15 kb windows of each telomere. Plots are as described in Figure 1D. (G) Single-read plots from long-read Nanopore sequencing of SIR3-M.ECOGII (JRY13027) and sir3-bah∆-M.ECOGII (JRY13438) at two representative telomeres (1 L and 2 L). Shown are 10 kb windows of each telomere.
The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Uncropped protein immunoblot of Sir3 mutants.    Figure 4E). This result suggested that binding of Sir3 at silencers and promoters was sufficient for partial repression and preceded its spread over the entirety of both loci.

Discussion
In this study, we developed a new method to study the process of recruitment and spread of the S. cerevisiae heterochromatin protein Sir3 in living cells with a resolution approximating the frequency of single A-T basepairs. We created a fusion protein between a key structural protein of heterochromatin, Sir3, and the bacterial adenine methyltransferase M.EcoGII that retained function and activity of  Aggregate methylation results at a control region on chromosome III to show background levels of methylation (top row), at HML (middle row) and HMR (bottom row) from long-read Nanopore sequencing of the same strains in A. The two colors for each genotype correspond to two biological replicates. Plots are as described in D. (E) RT-qPCR of CHA1 and OCA4 mRNA normalized to ACT1 mRNA in the same strains as C. (F) Aggregate methylation results at three representative telomeres (6 R, 15 R, and 10 R) from long-read Nanopore sequencing of the same strains as A. The two colors for each genotype correspond to two biological replicates. Shown are 15 kb windows of each telomere. Plots are as described in Figure 1D.
The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Uncropped Protein Immunoblot of Sir3-M.EcoGII overexpression strains.  each. We used DNA methylation as a read-out for Sir3 occupancy on chromatin. Long-read Nanopore sequencing allowed us to distinguish directly between methylated and unmethylated adenine and transcended the limitations of earlier studies imposed by repetitive regions common at telomeres. The methylation by Sir3-M.EcoGII at HML and HMR was stronger and had a larger footprint than its occupancy as judged by ChIP-seq, suggesting that our method captured transient contacts of Sir3 with chromatin that ChIP-seq could not. This result reinforced the idea that protein-chromatin interactions are dynamic, even for a feature like heterochromatin, commonly thought of as 'stable'. We also harnessed the power of single basepair resolution afforded by Nanopore sequencing to distinguish between recruitment and spread of Sir3 by studying a mutant of Sir3 whose distribution and binding profile had not yet been characterized, sir3-bah∆. This mutant cannot bind to nucleosomes and loses transcriptional silencing at HML and HMR (Buchberger et al., 2008;Gotta et al., 1998;Onishi et al., 2007). The mutant sir3-bah∆-M.EcoGII protein was still recruited to silencers at HML, HMR, and to telomeres where it methylated local adenines. However, the mutant did not spread beyond those recruitment sites, unlike wild-type Sir3-M.EcoGII. Our findings supported that recruitment and spread were separate processes that involved different interactions between Sir3 and other proteins at silenced loci and telomeres. The separation of nucleation and spread   mechanisms are also seen in studies of S. pombe heterochromatin. In S. pombe, transcriptional silencing of mating-type loci involves recruitment of histone deacetylases, histone methyltransferases, and the chromatin-binding factor Swi6/HP1 to sequence-specific nucleation sites followed by spreading of these factors over mating-type gene bodies to silence them. Most notably, spreading, but not nucleation, requires a 'read-write' mechanism in which the complexes that methylate histone H3 on lysine 9 (H3K9) also recognize that methylation mark (Akoury et al., 2019;Al-Sady et al., 2013;Ivanova et al., 1998;Zhang et al., 2008). Without this 'read-write' mechanism in place, heterochromatin factors are recruited, but they do not spread beyond their nucleation sites (Noma et al., 2004;Yamada et al., 2005;Zhang et al., 2008). The ability to unambiguously map long reads to telomeres allowed us to challenge historic conclusions about Sir3 dosage-driven heterochromatin spreading. Overexpressing Sir3-M.EcoGII increased the methylation signal where there already was methylation at endogenous levels of expression, including a large increase in methylation between HML and telomere 3 L. However, the boundaries where methylation dropped to sir3∆::M.ECOGII levels at HML, HMR, and telomeres mainly remained fixed in the two conditions. This result suggested that overexpression of Sir3 was not sufficient for spreading past most wild-type boundaries. The original idea that overexpression of Sir3 results in its further spread relied on low-resolution RT-PCR at two telomeres, 5 R and 6 R (Hecht et al., 1996;Renauld et al., 1993;Strahl-Bolsinger et al., 1997). In our genome-wide analysis, telomere 6 R was an exception, not the rule, as most telomeres did not show spreading of Sir3 upon its overexpression. We were not able to reproduce the result at telomere 5 R (Hecht et al., 1996).
Two results suggested that repression of HML and HMR did not require that Sir3 occupy the entire locus to the level seen in wild-type cells. During the temperature switch time course, methylation by sir3-8-M.EcoGII appeared first and most strongly at promoters and silencers of HML and HMR, with little or no detectable methylation between these sites. Yet, partial transcriptional repression was already evident at both loci within 30 min. The gradual repression that appeared during the time course was consistent with single-cell studies that show gradual tuning down of transcription during silencing establishment (Goodnight and Rine, 2020). The level of repression achieved during the time course was commensurate with that expected within the first cell cycle following restoration of Sir3 function (Goodnight and Rine, 2020). We also found that sir3-bah∆, a nucleosome-binding mutant that was recruited to silencers and promoters but could not bind outside of them, achieved some repression at both HML and HMR. Repression was stronger at HMR than at HML in the bah∆ mutant, perhaps because HMR is smaller and less dependent on spreading. The time course and bah∆ results together suggested a difference between repression of transcription per se and the stability of silencing ultimately achieved by the SIR complex binding over the entirety of HML and HMR.
We focused our efforts on the heterochromatin protein Sir3 in S. cerevisiae, but this method may have broad utility. By relaxing the requirement for stable binding to detect interaction of a protein with DNA or chromatin, the method could find binding sites of transcription factors and/or chromatinbinding proteins that elude detection by ChIP-seq. Further, with an inducible M.EcoGII fusion protein, one could track processes over time like the spread of heterochromatin proteins during X-chromosome inactivation, the movement of cohesin and condensin complexes along chromosomes during chromosome pairing and condensation, and the homology search during homologous recombination. Improvements in Nanopore technology and modified base-calling software will likely extend the method's utility to other processes that have directional movement along chromosomes.

Strains
All strains in this study were derived from W303 (Supplementary file 1). For biological replicates, independent cultures were started from the same strain or isogenic strains as indicated by strain numbers in figure legends. M.ECOGII integrations were created by one-step integration of a PCRamplified M.ECOGII-natMX cassette from pJR3525 using the primers listed in Supplementary file 3. Deletion of the BAH domain of SIR3 was done using CRISPR-Cas9 gene editing using the guide RNA and repair template listed in Supplementary file 3. The guide RNA target and nontarget strands were integrated into a single guide RNA dropout-Cas9 expression plasmid (pJR3428, Brothers and Rine, 2019) by Golden Gate cloning, using the restriction enzyme BsmBI as described in Lee et al., 2015. The repair template was made by annealing oligos described in Supplementary file 3 and extending the 3′ ends using Phusion Polymerase (New England Biolabs, Beverly, MA). SIR3 overexpression strain and its control strain were created by transformation and maintenance of 2 µm plasmids pJR3526 and YEp24, respectively.

Plasmids
Plasmids used in this study are listed in Supplementary file 2. The M.ECOGII-NatMX6 tagging plasmid (pJR3525) was made using standard Gibson cloning into pFA6a-natMX6 (Hentges et al., 2005). The codon-optimized M.ECOGII ORF with homology to the vector backbone was on a gene block (Supplementary file 3) from Integrated DNA Technologies (IDT, Coralville, IA) and was inserted into pFA6a-natMX6 linearized by PCR. The SIR3p-SIR3-M.ECOGII overexpression plasmid (pJR3526) was made using standard Gibson cloning into YEp24, a 2μ yeast expression plasmid carrying a URA3 selectable marker (Botstein et al., 1979). The sequences for each plasmid are available in Supplementary Materials as fasta and. dna files.

ChIP-seq Sample collection
Strains were grown to mid-log phase in YPD at 30°C. Approximately 10 9 cells were collected, washed, and fixed for 15 min at 30°C in a final concentration of 1% formaldehyde. Fixation was quenched with a final concentration of 300 mM glycine for 10 min at 30°C. Cells were washed 1× with PBS and 2× with FA lysis buffer (50 mM HEPES, pH7.5, 150 mM NaCl, 1 mM EDTA, 1% [vol/vol] Triton X-100, 0.1% [wt/vol] sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS]) before flash freezing pellets in a 2-ml screw-cap tube. Pellets were resuspended in 1 ml FA lysis buffer and 500 μl of 0.5 mm zirconium ceramic beads (BioSpec Products, Bartlesville, OK, Cat # 11079105z) were added. Resuspended cells were bead beat with in a FastPrep-24 (MP Biomedicals, Burlingame, CA) at 5.5 amplitude, four cycles of 40 s ON/2 min on ice. Each tube was punctured at the bottom with a hot 20-G needle and placed into a new 1.5 ml tube, and sample was spun out of the tube into the new tube by spinning at 150 × g for 1 min. The sample was moved into a 15-ml Bioruptor sonication conical tube with 100 μl of Bioruptor sonication beads (Diagenode, Denville, NJ, Cat # C01020031) and sonicated using the Bioruptor Pico (Diagenode) for 10 cycles of 30 s ON/30 s OFF.

Immunoprecipitation
The sonicated extract was moved into a new 1.5 ml tube and spun at 16,000 × g for 15 min at 4°C. The supernatant was moved into a new 1.5 ml tube and adjusted to 1 ml volume with FA lysis buffer. 50 μl of sample was taken aside as input, and then 25 μl of 20 mg/ml BSA and 5 μl of mouse anti-V5 antibody (Invitrogen, Waltham, MA, Cat # R960-25) or 10 μl of polyclonal rabbit anti-Sir3 antibody (from N. Dhillon & R. Kamakaka) was added to the rest of the sample and rotated overnight at 4°C. 50 μl of Protein A magnetic Dynabeads (Thermo Fisher, Waltham, MA, Cat # 10,002D) were added to the sample and rotated at 4°C for 1 hr. Magnetic beads were immobilized using a magnetic rack and washed by resuspension in 1 ml of various buffers in the following order: FA lysis buffer + 0.05% Tween-20, FA lysis buffer + 0.05% Tween-20 + 0.25 mM NaCl, ChIP wash buffer (10 mM Tris, pH 8.0, 0.25 M LiCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.5% sodium deoxycholate, 0.05% Tween-20), and TE (10 mM Tris, pH 8.0, 1 mM EDTA) + 0.05% Tween-20. The washed beads were resuspended in 130 μl of ChIP elution buffer (10 mM Tris, pH 7.5, 1 mM EDTA, 1% SDS) and incubated at 65°C shaking at 900 rpm overnight. The next day, 2.5 μl of 10 mg/ml Proteinase K (New England Biolabs, Ipswich, MA, Cat # P8107S) and 2.5 μl of 10 mg/ml RNase A (Thermo Fisher, Cat # EN0531) were added to the elution and incubated at 42°C for 2 hr. Beads were immobilized on a magnetic rack and the supernatant containing the desired DNA to be sequenced was taken and purified using 1× (vol/vol) SPRI Select magnetic beads (Beckman Coulter, Brea, CA, Cat # B23317) according to the manufacturer's instructions.

Library preparation and sequencing
Samples were prepared for sequencing using NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs, Cat # E7645) according to the manufacturer's instructions. Samples were multiplexed using NEBNext Multiplex Oligos for Illumina (New England Biolabs, Cat # E7335/E7500). Library-prepped samples were sequenced on a MiniSeq System (Illumina, San Diego, CA).

DIP-seq DNA extraction
Cells were grown to mid-log phase in YPD, Complete Supplement Mixture (CSM), or CSM without Uracil (Sunrise Science Products, Knoxville, TN) at 30°C. Approximately 10 9 cells were pelleted by centrifugation at 3200 × g for 2 min, washed with 1 ml of water, moved to a 2-ml screw-cap tube, and flash frozen. Cells were resuspended in 400 μl of Triton SDS Lysis Buffer (10 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 2% Triton X-100, 1% SDS), and 400 μl of phenol:chloroform:isoamyl alcohol 25:24:1 and 300 μl of 0.5 mm zirconium ceramic beads (BioSpec Products, Cat # 11079105z) were added to the resuspension. Cells were lysed by bead beating with in a FastPrep-24 (MP Biomedicals) at 5.5 amplitude, 4 cycles of 40 s ON/2 min on ice. The aqueous and organic phases were separated by centrifugation at 21,000 × g for 5 min, and the aqueous phase was moved to a new 1.5 ml tube. 400 μl of chloroform was added, vortexed at top speed for ~10 s, and spun down at 21,000 × g for 5 min to separate the aqueous and organic phases. The aqueous phase was moved to a new 1.5 ml tube, and 1 ml of 100% ethanol was added to precipitate nucleic acids. The sample was incubated at 4°C for 10-15 min and then spun down at 21,000 × g for 2 min to pellet the precipitated nucleic acids. Supernatant was discarded, the pellet was air-dried, and then the pellet was resuspended in 400 μl of TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) + 4 μl of 10 mg/ml RNase A (Thermo Fisher, Cat # EN0531) and incubated at 37°C for 1 hr. 1 ml of 100% ethanol + 10 μl of 4 M ammonium acetate were added to the RNase solution and incubated at 4°C for 10-15 min to precipitate DNA. The precipitate was pelleted by centrifugation at 21,000 × g for 2 min, washed 1× with 70% ethanol, air-dried, and resuspended in 150-300 μl of water.

Sonication
DNA concentration was measured using Qubit dsDNA HS reagents (Invitrogen, Cat #Q32854), and 6 μg of DNA was diluted to 20 ng/µl in 300 μl of water in 1.5 ml Bioruptor Pico Microtubes for sonication (Diagenode, Cat # C30010016). DNA was sonicated using a Bioruptor Pico (Diagenode) for 18 cycles of 15 s ON/90 s OFF. The sonicated DNA was moved to a new 1.5 ml tube. m 6 A IP DNA was denatured by incubating at 95°C for 10 min and then immediately placed on ice for 5 min. 200 μl of cold water and 500 μl of cold 5× DIP buffer (50 mM NaPO 4 , pH 7.0, 700 mM NaCl, 0.25% Triton X-100) were added to bring the volume up to 1 ml. 50 μl was taken aside as input. 25 μl of 20 mg/ml BSA and 1.8 μg of antibody (Synaptic Systems rabbit anti-m 6 A, Cat 202-003) were added to the rest of the sample and rotated overnight at 4°C. 50 μl of Protein A magnetic Dynabeads (Thermo Fisher, Cat # 10,002D) were added to the sample and rotated at 4°C for 1 hr. Magnetic beads were immobilized using a magnetic rack and washed by resuspension and rotation for 5 min at 4°C in 1 ml of various cold buffers in the following order: 2× with 1× DIP buffer + 0.05% Tween-20, 1× with 1× DIP buffer. For elution, beads were resuspended in 190 μl of DIP digestion buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 0.5% SDS) + 10 μl of 10 mg/ml Proteinase K (New England Biolabs, Cat # P8107S). DIP digestion buffer was added to input samples up to 200 μl. Both the input and IP samples were incubated at 50°C for 2 hr and then cleaned up using the Qiaquick PCR Purification Kit (Qiagen, Hilden, Germany, Cat # 28104) and eluted in 35 μl of water.

Library preparation and sequencing
Samples were prepared for sequencing using Accel-NGS 1S Plus DNA Library Kit (Swift Biosciences, Ann Arbor, MI, Cat # 10024) according to the manufacturer's instructions. Samples were multiplexed using Swift Single Indexing Primers Set A (Swift Biosciences, Cat # X6024). Library-prepped samples were sequenced on a MiniSeq System (Illumina).

Analysis
Analysis was done as described in the section on ChIP-seq above.

Nanopore sequencing DNA extraction
Cells were grown to mid-log phase in YPD, CSM (Sunrise Science Products), or CSM without Uracil at 30°C. Approximately 10 8 cells were pelleted, washed with 1 ml of water, and pellets were flash frozen. gDNA was extracted using the YeaStar Genomic DNA Kit (Genesee Scientific, San Diego, CA, Cat #11-323) according to the manufacturer's 'Protocol 1'. Specifically, thawed cell pellets were resuspended in 240 μl of YD digestion buffer + 10 μl R-Zymolyase and incubated at 30°C for 1 hr. 240 μl of YD Lysis buffer was added to the solution and vortexed at top speed for 15 s. 500 μl of chloroform was added to the solution and vortexed at top speed for 10 s and then inverted 10 times. The aqueous and organic phases were separated by centrifugation at 10,000 × g for 2 min, and the aqueous phase was equally separated into two ZymoSpin columns. ZymoSpin columns were spun at 10,000 × g, washed 2× with 300 μl of DNA wash buffer, and DNA was eluted from each column with 75 μl of water. Eluates were combined. DNA was sheared to ~15-20 kb by spinning through a Covaris g-TUBE (Covaris Inc, Woburn, MA, Cat # 520079) at 4200 rpm for 1 min, and repeating 1× with the tube flipped the other way in an Eppendorf Centrifuge 5424, according to the Covaris protocol. DNA was purified and concentrated using 1× (vol/vol) SPRI Select beads (Beckman Coulter, Cat # B23317) and eluted in 50 μl of water according to the manufacturer's instructions. DNA concentration was measured using Qubit dsDNA HS reagents (Invitrogen, Cat # Q32854).
Results were aggregated into. bed files using 'megalodon_extras aggregate run', and these files were used for aggregate Nanopore plots. Before plotting, aggregated data were filtered to include only adenines with at least 10× coverage, and lines were smoothed using base R loess() function with enp. target = 100 and weighted by the coverage at each position.
The per-read database from Megalodon was converted into a.txt file using 'megalodon_extras per_read_text modified_bases'. For ease of use in RStudio, the data for each chromosome were extracted into its own. txt file using custom bash and awk scripts (Source code 4) and these files were used for single-read Nanopore plots. The probabilities output by Megalodon were made binary by calling adenines with a > 0.8 probability of being methylated as 'm6A' and all others 'A'.
The R scripts ( as. html files) used to create each figure can be found in Source code 5.

Limitations
This method showed possible limitations for some contexts: (1) The expression level of the fusion protein could increase the levels of background methylation. We found this to be true with the Sir2-M. EcoGII fusion protein, which is likely expressed at a higher level than Sir3-M.EcoGII due to the higher level of endogenous Sir2 expression. The level of methylation by Sir2-M.EcoGII in heterochromatin regions was higher than by Sir3-M.EcoGII, and background levels of methylation outside of heterochromatin regions were also higher than by Sir3-M.EcoGII. Importantly, the signal at heterochromatin was evident above even this raised background methylation.
(2) Methylation at the level of single reads was variable and spotty, possibly due to at least two contributors. There may be occupancies that are too transient to allow methylation. Secondly, computational limitations for calling modified adenines without a guiding sequence motif meant that lower-confidence (probably <0.8) m 6 A calls were not considered methylated. At present, qualitative conclusions based on single-read data can be made with confidence, but as Nanopore technology improves, single-read data will become more amenable to statistical and spatial analysis.
(3) Because this method can capture transient interactions better than methods like ChIP-seq it may overestimate degrees of occupancy unless combined with DIP-or ChIP-seq.

Reverse Transcription and Quantitative PCR (RT-qPCR) RNA extraction
Cells were grown to mid-log phase in YPD, CSM (Sunrise Science Products), or CSM-Uracil at 30°C, and RNA was extracted using the Qiagen RNeasy kit (Qiagen, Cat # 74104) according to the manufacturer's instructions for purification of total RNA from yeast. Briefly, ~6 × 10 7 cells were resuspended in 600 μl of buffer RLT, 500 μl of 0.5 mm zirconium ceramic beads (BioSpec Products, Cat # 11079105z) were added, and cells were lysed by bead beating with in a FastPrep-24 (MP Biomedicals) at 5.5 amplitude, 3 cycles of 40 s ON/2 min on ice. Cells were pelleted by spinning at 21,000 × g for 2 min, and the supernatant was moved to a new tube. One volume of 70% ethanol was added to the supernatant and the sample was spun through an RNeasy spin column. The column was washed with 350 μl of buffer RW1, then 10 μl of DNase + 70 μl of buffer RDD (Qiagen, Cat # 79256) were added to the column and incubated for 15 min at room temperature. 500 μl of buffer RW1 was added and spun through the column. The column was then washed with 500 μl of buffer RPE 2×, and RNA was eluted with 80-150 μl of RNase-free water.

RT-qPCR
Complementary DNA was synthesized using the SuperScript III First-Strand Synthesis System (Invitrogen, Cat # 18080051) and oligo(dT) primers according to the manufacturer's protocols. Quantitative PCR of complementary DNA was performed using the DyNAmo HS SYBR Green kit (Thermo Fisher, Cat # F410L) on an Mx3000P machine (Stratagene, La Jolla, CA) using the primers listed in Supplementary file 3. Standard curves were generated using a tenfold dilution series of one of the prepared samples.

Author contributions
• Supplementary file 3. Oligonucleotides used in this study.
• Source code 1. Creation of bedgraph files from fastq files.
• Source code 2. Normalization of bedgraph files to genome-wide median.
• Source code 3. Extract nanopore read IDs corresponding to each barcode from sequencing summary file.
• Source code 4. Extract single read data corresponding to a given chromosome.
• Source code 5. R scripts used to construct nanopore data into figures.

Data availability
All strains (Supplementary File 1) and plasmids (Supplementary File 2) are available upon request. Sequencing data is available in GEO under the SuperSeries GSE190137. ChIP-seq and DIP-seq data are under accession code GSE189038 in the SuperSeries. Nanopore data are under accession code GSE190136 in the SuperSeries.
The following dataset was generated: Author (